From June 13th to 24th 2025, Iran and Israel went to war.
Like many on both sides of the conflict (I live in Jerusalem), my life was suddenly upended by it.
I also lived through the Iranian ballistic missile attacks of 2024 which occurred in two waves: The first on April 13th (2024) and a much larger attack (codenamed Operation True Promise II) on October 1st. During the latter, Iran launched approximately 200 ballistic missiles.
This experience has formed a few key beliefs:
Read any consular bulletin or public safety alert given to those sheltering during times of conflict and "keep abreast of local news developments" will be very close to the top recommendation, if not the very first.
The problem is that today's 24/7 news cycle causes a seriously degraded noise-to-signal ratio:
- During wartime, public broadcasters in Israel - as elsewhere - begin uninterrupted coverage of the crisis/war/emergency
- Some content contains immediately actionable guidance for civilians ("seek shelter now!"), but most consists of analysis and speculation
- Watching continuously frays already tense nerves
Yet many feel they have no choice but to stay tuned because they don't want to miss potentially life-saving information.
A thought that has crossed my mind many times during Israel's crises, when worn out and exhausted: "I want to sleep, but I don't know when it's safe to turn this thing off!"
The cornerstone of preparedness efforts involves being prepared for disaster.
Israel now operates a multi-layered civilian alerting system:
- 📱 Home Front Command (Pikud HaOref) provides front-line alerting via smartphone applications and website. Third-party apps exist (including systems that scrape and syndicate alerts to channels like X and Telegram), but they essentially replicate and extend the digital channel
- 📞 Cell Broadcast Technology - Implemented in 2025 for non-data-enabled phones (used by a sizeable proportion of Israel's ultra-Orthodox population). This uses a protocol layered directly into the GSM stack, meaning alerts can be received without cellular data (helpful for roaming users)
- 🔊 Nationwide Siren System issues audible air raid sirens
While this system is mature and well-designed, I believe a slightly preemptive alerting system would have independent value.
During the recent conflict with Iran, some salvos used hypersonic munitions that were sent in the middle of the night.
A sizeable portion of the Israeli population lacks a private shelter. Given the extremely fast speed at which these munitions travel, civilians needed to reach the nearest shelter within 5 minutes after being woken by an alert in the middle of the night.
Since the first round of hostilities with Iran in 2024, I've been experimenting with different ways to set up a monitoring system (enhanced with AI or not) that provides targeted alerts based on open-source information.
I noticed that both the October 2024 attacks and those in 2025 were preceded by news coverage containing clear indications of imminent hostilities:
- During the 2025 conflict, news that the US had begun evacuating diplomats from the Middle East broke around lunchtime the day before Israel launched its campaign
- Similar patterns emerged in previous conflicts
This raises an important question: What keyword groups could be leveraged to provide alerts based on open-source signals that indicate hostilities may break out in the next 12-hour window?
Furthermore:
During the Iran-Israel war, it became apparent that preparations for imminent missile salvos were being communicated—often via IRGC propaganda channels and spokespeople—shortly before the events took place.
Selectively filtered alerting based on these communications could provide citizens and interested parties with precious additional preparation time.
My plan is structured around three key objectives:
-
🔎 Retrospective Analysis - Examine information accessible through open channels (particularly X and Telegram) that may have provided forward notice, even if only by a few minutes, ahead of emergency alerting apps
-
⏱️ Timestamp Comparison - Compare timestamps against those of issued emergency alerts and broadcasts to identify which sources may have provided useful OSINT clues to those in the region
-
🧩 Pattern Recognition - Work from signal to search, hypothesizing which keyword searches and monitoring strategies may prove fruitful in future conflicts